It's a bit of a misconception that the shuttle used S-turns to slow down. To quote the Entry, TAEM, and Approach/Landing Guidance Workbook by the United Space Alliance:
The next time you hear someone talk about the shuttle doing roll reversals to bleed off energy, do not listen. The shuttle does roll reversals because it has a very small alpha envelope.
What it's saying is that the ideal reentry strategy is to simply point the nose at the landing site and control how fast you're dissipating energy by controlling angle of attack (alpha). A higher alpha will result in more drag, and it's easy to make quick adjustments to get the level of drag you need.
However, in order keep the orbiter both controllable and protected by the thermal protection system, it had to be flown at a particular alpha (40° for most of entry) with no more than 3° of variation. This meant they had to use different means to control drag, and the way they did it was by controlling the lift vector.
A steeper bank angle meant that the lift vector would be more sideways than up, which would cause the shuttle to descend quicker, and, as you descend, the air gets thicker, which causes more drag. A shallower bank angle will slow your descent and keep you in thinner air for longer, which minimizes your drag.
But there's a problem with using bank... it starts to turn you off course. So the solution is to use roll reversals (aka s-turns) to keep you pointed towards the landing site.
Now, to sort of answer the literal question you asked (how much entry speed was bled off during the s-turns): most of it. The first roll would occur at nearly orbital velocity, and entry guidance ended at 2500 feet per second (Earth-relative velocity). After entry guidance ended, TAEM (Terminal Area Energy Management) guidance began, which primarily used alpha for energy management. Once the orbiter was subsonic, the speedbrake was also used for energy management.